How to safely stop a `chkdsk /r /f C:` on Windows 10?

Question

I started a chkdsk /r /f C: on Windows 10, but now that it's running I want to cancel it. Just powering down the computer risks corruption, so how can I safely abort it?

Ctrl+C isn't an option: I'm running chkdsk /r /f on the drive that has Windows installed. This cannot be done while Windows is running, but only during startup (outside of CMD). This doesn't respond to ctrl+c.

Note: the linked duplicate question is NOT the same. That question is about running chkdsk without parameters, and that is safe because it runs in read-only mode. The /r /f flags causes chkdsk to run in read-write mode, so then it's not generally safe to just kill the process. It needs to be terminated gracefully. Some implementations of fsck (linux equivalent) can be stopped gracefully - even in repair mode - so theoretically it should definitely be possible to safely stop a chkdsk procedure. The main question is: did the Windows devs actually implement a graceful cancellation procedure, and if so how do I trigger it?

Answer 1

Edit: As noted in the comments, there is no safe option (but you could use more or less riskier moments).

If it is running in the pre-startup and you are past Stage 3, meaning it is on 4 of 5 or 5 of 5, then you can just restart your machine. I have done it on stage 5 of 5, which is the stage of checking the free space, and it started up just fine. If you restart your machine during stages 1 to 3, you run the risk of losing data.

Source: Norman Picard

Answer 2

You can't stop chkdsk process once it started. The safe way is to wait until it completes. Stopping the computer during the check might lead to filesystem corruption.

Answer 3

Not sure since which version of Windows or of chkdsk itself this has been the case, but the Microsoft Learn page about it says

canceling or interrupting chkdsk should not leave the volume any more corrupt than it was before chkdsk was run

https://learn.microsoft.com/en-us/windows-server/administration/windows-commands/chkdsk?tabs=event-viewer

I trusted that,

• used Sysinternals' Process Explorer to first Suspend the chkdsk process (and listen for an end to all HDD noises) just for some added peace of mind, then
• opened another cmd window with elevated privileges and did a "chkdsk /x e:" (use your own driver letter) because I saw this suggested elsewhere, and /x:

Forces the volume to dismount first, if necessary. All open handles to the drive are invalidated.

• and finally pressed Del in Process Explorer to actually kill the original chkdsk process.

No big catastrophe ensued, the disk had a few kB of corruption here and there, but overall only in proportions I already expected because I hadn't been refreshing most of the data for over 1-1.5 decades. (But I also have a second copy I could restore anything from after comparing everything with Beyond Compare.)

Ideally don't stop chkdsk, even Microsoft don't seem so sure of their own answer as you can see from their choice of words: "should not" leave the volume any more corrupted. But if risk/benefit analysis forces you to, I would say try to be as unabrupt about it as possible, try to separate the process into some stages like pause/suspend/dismount/etc. (for example now I'm thinking I should've tried a Windows shutdown after suspending the process, instead of killing it myself, maybe that would take more care to "deactivate more stuff" in a civilized way and somehow create better conditions for finally killing the process and also getting the HDD's own controller do some cleanup of its own before being powered down).

Answer 4

Do not do this after /r command. I have followed the advice of stopping it and it caused this issue. Now I am doing chkdsk D: /r /x.

Answer 5

Don't know if safe, but I successfully stopped CHKDSK with the following line command on an elevated cmd window:
taskkill /F PID nnnn
Where nnnn is the Process ID for the running CHKDSK.
To obtain the PID, open the Task Manager and look at CHKDSK -> right click - Go to details
To run an elevated cmd window, go to taskbar, search -> cmd (or command prompt) - right click - Run as administrator.
I did it in the first stage, when searching for errors, so i hope it was pretty safe.
In a stage where it is writing fixes could be risky i think.
My unit already reported unspecified "errors" before this in a previous read-only run, so I started with /F fix option, but then I decided to stop to try better techniques.

Answer 6

Here's my output from chkdsk

File record segment 36383 is unreadable.
 1 percent complete. (36384 of 247328 file records processed)
File record segment 36384 is unreadable.
 1 percent complete. (36385 of 247328 file records processed)
File record segment 36385 is unreadable.
 1 percent complete. (36386 of 247328 file records processed)
File record segment 36386 is unreadable.
 1 percent complete. (36387 of 247328 file records processed)
File record segment 36387 is unreadable.
 1 percent complete. (36388 of 247328 file records processed)
File record segment 36388 is unreadable.
 1 percent complete. (36389 of 247328 file records processed)
File record segment 36389 is unreadable.
 1 percent complete. (36390 of 247328 file records processed)

I think it's safe to kill the PID in my case ;)

Answer 7

I got really impatient, since it was counting like 300k from 2000k+ (I don't know what was it counting, I don't know about OS), and hit Alt+F4 in the cmd window. I believe this is smoother than killing it from the task manager.

Answer 8

Alt/F4 was my exit. Immediately did a W10 disk scan. No errors found. Glary Utilities has a disk check feature. It will let you stop any check. Did a file check with that and then a scan for bad sectors. All seems well at this point.